Move to Tumblr

After a short conversation with a friend last night, I decided to move my blog to Tumblr. He said "blogspot is like hotmail". Well, I guess he's right. Guys, this is my new address: http://annezhou.tumblr.com/. Check it out!

205 trillion options for your learning!

Today's post will start from an interesting experiment conducted by a Human Computer Interaction professor at Carnegie Mellon University. In his experiment, he gave both students and educators a math problem in three different formats: story, equation and the narrative of the equation.

• Story: Tom plan to share his 26 bottles of beer with 4 friends. In order to make everyone get 2, how many bottles he should leave for himself? 
• Equation: (26-x)/4=2

Now guess which student group gets higher correction percentage? 

Unfortunately, I pick up the wrong option as most of the educators do. :( Different from educators' judgement (equation group may be easier for students to understand), story telling gets higher correction percentage! He does this experiment in both high schools and universities across different countries and gets the same result. (The difference between storytelling and equation may be less significant in college level students) 

Prof. Kenneth Koedinger, presenter of the seminar considers this as a "linguistic phenomenon". In his point of view, the equation is an "information chunk". For those who know the rules (like the educator), it is easier to understand; but if you don't know the chunk, you cannot solve the problem immediately (like the students). He uses an analogy for the students' difficulty in this situation: imagine you're playing chess. It is hard for someone who doesn't understand the rule to know how to get there. That's exactly what happened here--the gap between the educators and the students. In order to overcome that gap, he suggests a "data-driven" learning method. He believes that educational design is better based on accurate cognitive models of student knowledge and learning. And accurate cognitive models are best built by combining constraints from computational modeling and from statistical models of student performance data. 

I'm also amused by his PowerPoint about "instructional complexity": basically, there are 205 trillion options when you decide to design your instructions. So educators, do NOT expect to pick up the "right" method from all those options. Time to do your data mining job!

P.S: Last week, Stanford Computer Science department announced CS+X combination major option. For now, X could be English or Music. I think this is a perfect interpretation of the "linguistic phenomenon". 

Seminar at Stanford HCI

Last Friday, I went to a GREAT seminar hosted by Stanford Human-Computer Interactive Group (HCI). The presenter Mira Doutcheva is  a senior research scientist at Adobe Systems. Her work in Adobe (and also this seminar) is focused on: 1) how to lower the users' barriers with the help of creativity support tools and 2) how to help designers and developers to build these tools to support the information need. Check more detailed information here.

The tool called "Pixeltone" makes me extremely interested. Check its introduction video here and see how cool it is!

Of course it makes those complicated steps to accomplish a task become much easier with help of audio instruction.  Assuming you are an experienced photoshop user, you have to go through AT LEAST four steps ("edit-color-red-apply") to change the color of a shirt.  But with Pixeltone, it only takes two ("define a shirt"-"change it red"), even for a beginner. But I see its great potential to get over the "natural language" problem, which is a big challenge for most of the designers. You cannot expect your users to know what "rinse" is for the first time, and it may take them so long to figure out what does that mean. (Unfortunately, "rinse" is not the only "Designer's Language" they need to figure out).

So the mechanism of PixelTone is that it analyzes the users' data by breaking the syntax of the sentence. For example: "Make the shirt bright" will be "Verb+n.+adj." in the database. The computer will search the database and find the correct action next. That means you don't have to find the ONLY correct path ("edit-color-red-apply"). Just leave the work to your computer. But remember: Be nice! Don't expect your laptop know what you mean by "make it heavenly" :)